The problem

At Railroad19 I built dashboard interfaces for a major media company's ad-sales teams, turning complex order data into clear visualizations. Those dashboards helped teams spot and act on $6.6M in mismarked orders – and a lot of that work happens in tables.

As the app grew, each new table type picked up its own row component. The rows became unwieldy: the same display logic, copied with small variations, and every new table meant solving the same problems again.

What I did

I redesigned the rows as dynamic, reusable components that could be selected dynamically via JSON config. The display logic lives in one place, with the ability to pass in custom logic where needed, so other engineers can build a new table type on top of the same foundation.

The result

The approach became a pattern other engineers on the team adopted, and new tables no longer needed to duplicate display logic for common cell types.